Saturday, November 27, 2010

Electrical Fault Detection in Power Systems by ANN structures

  MICHAŁ SZEWCZYK ADRIAN HALINKA Institute of Power Systems and Control  Silesian University of Technology ul. B. Krzywoustego 2 POLAND

Abstract: Nowadays used power system protections are worked out to get the properties of digital technology. This improves the reliability and functionality of protection devices by increasing the number of possible information received from protected object,  self testing, monitoring and logging the events. However the decision parts of such systems are still based on commonly used principles defined already for analog protection devices. The exceeding of threshold value cause the generation of respectively logical signals (mainly “0” and “1”) which are adequately interpreted as “normal” and “faulty” conditions of protected object. Such approach to making a decision, considerably narrow down the area of recognized states. One of the methods to define of decision system, not based on the above mentioned principle, is the use of artificial neural networks (ANN). In the paper a model of decision system based on ANN, will be shown. As protected object the generator-transformer unit has been taking into consideration. The range of detected faults are initially narrows to faults of electromagnetic character (three-phase, two-phase, two-phase to earth and one-phase faults) within the generator – unit transformer – high voltage transmission line configuration.

Keywords: Artificial neural network, power system protection, electrical fault detection, generator-transformer unit.


No comments:

Post a Comment